flowMap

flowMap maps and quantifies similarity of cell populations across flow cytometry (FCM) samples by applying the Friedman-Rafsky (FR) nonparametric test statistic to assess equivalence of multivariate distributions.

Key Features:

• Friedman-Rafsky statistic: Uses the Friedman-Rafsky nonparametric test statistic to measure equivalence of multivariate distributions from FCM fluorescence data.
• Distribution-based similarity: Quantifies similarity between cell populations based on fluorescence distribution characteristics such as shape, size, and position in multidimensional feature space.
• Simulation of variation: Simulates biological and technical variations commonly observed in FCM data to evaluate method performance.
• Robustness to shifts and proportions: Identifies equivalent cell populations despite proportion differences and modest position shifts.
• Gating error detection: Detects inappropriate splitting or merging of cell populations during gating.
• Performance comparison: Demonstrated improved discrimination of equivalent versus nonequivalent populations compared with the symmetric Kullback-Leibler divergence measure.
• Distance metric for matching: Serves as a distance metric to match manually gated cell populations across multiple FCM samples.
• Benchmark performance: Reported an F-measure of 0.88 on 30 FCM samples from the FlowCAP benchmark data set.

Scientific Applications:

• Biomarker identification: Supporting identification of cellular biomarkers that distinguish normal physiological responses from disease states by comparing population distributions.
• Marker expression testing: Determining whether expression of a cellular marker is statistically different between two cell populations.
• Phenotype discovery: Aiding detection of potential new cellular phenotypes through equivalence testing of multivariate distributions.
• Gating validation: Validating manual gating results by detecting incorrect splitting or merging of populations.
• Cross-sample matching: Matching corresponding cell populations across multiple FCM samples for comparative analyses.

Methodology:

Computes the Friedman-Rafsky nonparametric test statistic on multivariate FCM fluorescence data, uses simulated biological and technical variations to evaluate performance, and compares discrimination against the symmetric Kullback-Leibler divergence with performance reported by F-measure on FlowCAP benchmark samples.